CN112488417A - Power grid operation characteristic sensing method and system based on LBP and neural network - Google Patents

Power grid operation characteristic sensing method and system based on LBP and neural network Download PDF

Info

Publication number
CN112488417A
CN112488417A CN202011467995.4A CN202011467995A CN112488417A CN 112488417 A CN112488417 A CN 112488417A CN 202011467995 A CN202011467995 A CN 202011467995A CN 112488417 A CN112488417 A CN 112488417A
Authority
CN
China
Prior art keywords
power grid
data
situation
semantic information
neural network
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202011467995.4A
Other languages
Chinese (zh)
Other versions
CN112488417B (en
Inventor
赵奇
余璟
赵家庆
徐春雷
吕洋
丁宏恩
李春
田江
徐秀之
唐聪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Jiangsu Electric Power Co Ltd
Suzhou Power Supply Co of State Grid Jiangsu Electric Power Co Ltd
Original Assignee
State Grid Jiangsu Electric Power Co Ltd
Suzhou Power Supply Co of State Grid Jiangsu Electric Power Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by State Grid Jiangsu Electric Power Co Ltd, Suzhou Power Supply Co of State Grid Jiangsu Electric Power Co Ltd filed Critical State Grid Jiangsu Electric Power Co Ltd
Priority to CN202011467995.4A priority Critical patent/CN112488417B/en
Publication of CN112488417A publication Critical patent/CN112488417A/en
Application granted granted Critical
Publication of CN112488417B publication Critical patent/CN112488417B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/043—Architecture, e.g. interconnection topology based on fuzzy logic, fuzzy membership or fuzzy inference, e.g. adaptive neuro-fuzzy inference systems [ANFIS]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/02—Computing arrangements based on specific mathematical models using fuzzy logic
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computing Systems (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Biomedical Technology (AREA)
  • Marketing (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Biophysics (AREA)
  • Educational Administration (AREA)
  • Computational Linguistics (AREA)
  • Game Theory and Decision Science (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Automation & Control Theory (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Fuzzy Systems (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • Primary Health Care (AREA)

Abstract

The invention discloses a power grid operation characteristic sensing method and system based on LBP and a neural network. And converting the acquired fuzzy semantics into crisp numbers by adopting a triangular fuzzy number, and normalizing the data. And then, reducing the dimension of the acquired multidimensional data by adopting a local binary system (LBP) method, and accurately acquiring the characteristics of the power grid data. And finally, inputting the characteristic data and the corresponding power grid state information into a neural network for training to generate a power grid situation prediction model, so that the power grid situation can be accurately acquired. The LBP and neural network-based power grid operation characteristic sensing method and system can perform characteristic extraction on power grid multidimensional data, accurately control the power grid operation situation, provide support for power grid dispatching personnel to make dispatching strategies and provide guarantee for safe and reliable operation of the power grid.

Description

Power grid operation characteristic sensing method and system based on LBP and neural network
Technical Field
The invention belongs to the technical field of optimized operation of a power system, and relates to a power grid operation characteristic sensing method and system based on LBP and a neural network.
Background
With increasing concerns about fossil energy depletion and environmental deterioration, renewable energy is continuously accelerated to replace traditional energy, and the access of renewable energy to a power grid brings huge challenges to the power grid. How to fully utilize relevant data of power grid operation, accurately grasp the situation of the power grid, and improve the capability of the power grid in responding to the uncertainty of renewable energy sources becomes a key point of attention of the power grid.
At present, a method for extracting and analyzing the operating data characteristics of a power grid usually adopts a neural network or clustering method to process data, for example, a spectral clustering method is adopted to extract the characteristics of the icing process of the power transmission line in a document 'micro meteorological characteristic extraction model of the icing process of the power transmission line based on spectral clustering' (measurement and control technology, 2019, 7 months, Lixuan, Lipeng, Miao Aimin and the like), so as to realize the control of the whole process of the icing of the power line. The document "power grid load classification based on VMD and FCM clustering methods" (northeast electric power technology, 5 months in 2019, Jiazhida, Jiangfeng, Wanghaixin and the like) proposes a clustering method based on VMD and FCM to identify power grid load characteristics and realize classification of power grid loads. However, the above research is only analyzed from the grid side, and non-grid factors cannot be considered, so that it is difficult to ensure the accuracy of grid feature extraction. In addition, with the access of renewable energy, non-grid operation data also has a great influence on the operation of the grid, and the factor is often ignored in the conventional feature extraction, and the feature analysis is simply performed from the grid operation data level, so that the comprehensive prediction of the grid operation situation is difficult.
Disclosure of Invention
The invention aims to solve the problems of current power grid data feature extraction and situation accurate acquisition, provides a power grid operation data feature extraction and situation acquisition method of an LBP (local binary pattern) and a neural network, converts non-power grid operation data through triangular fuzzy numbers, performs feature extraction on the power grid data and the non-power grid data by adopting the LBP method, inputs a result into a BP (back propagation) neural network for training to obtain a power grid operation situation model, and finally realizes accurate acquisition of the power grid operation situation.
The invention adopts the following technical scheme:
a power grid operation characteristic sensing method and system based on LBP and a neural network comprises the following steps:
step 1: collecting power grid operation historical data, non-power grid operation historical data and power grid situation data; the power grid operation historical data comprises average voltage, average transmission power and power grid repair rate; the non-power grid operation historical data comprises historical data of weather, sunshine and environment at corresponding acquisition moments; the power grid situation data refer to power grid fault conditions and risk levels;
step 2: representing the non-power grid operation historical data and the power grid situation of the corresponding historical day through semantic information respectively, and converting the semantic information representing different risk levels of the non-power grid operation historical data range and the power grid situation into corresponding one-dimensional arrays, namely fragile numbers respectively;
and step 3: respectively carrying out normalization processing on the collected power grid operation data and the brittleness number converted in the step 2;
and 4, step 4: performing feature extraction on the power grid operation data and the brittleness number normalized in the step 3 to obtain feature data of power grid operation and feature data of non-power grid operation, and expressing the feature data in a numerical value form;
and 5: inputting the characteristic data extracted in the step (4) into a BP neural network for training to obtain a power grid situation prediction model;
step 6: and (5) collecting power grid operation data and non-power grid operation historical data, inputting the power grid operation data and the non-power grid operation historical data into the power grid situation prediction model trained in the step 5, and predicting the power grid situation.
The invention further comprises the following preferred embodiments:
in the step 1, collected power grid operation historical data is calendar historical data, the repair rate is the repair rate of the current day of the power grid, and the power grid situation is determined based on the repair rate of the current day of the power grid and the fault degree of the power grid;
the weather historical data refers to weather wind speed;
the sunshine history data refers to the intensity of solar radiation;
the environmental historical data specifically refers to historical air pollution indexes;
representing the power grid situation into five conditions of no risk, low risk, medium risk, high risk and fault state through semantic information;
and respectively representing the weather historical data, the sunshine historical data and the environment historical data as five conditions of high, common, low and low through semantic information.
The semantic information of the weather historical data is characterized in the following way:
when V belongs to [0,3.4), the weather semantic information is 'very low';
when V belongs to [3.4,8.0), the weather semantic information is 'low';
when V belongs to [8.0,10.7), the weather semantic information is 'general';
when V belongs to [10.8,13.8), the weather semantic information is 'high';
when V is larger than or equal to 13.8, the weather semantic information is 'very high';
wherein V is the current day wind speed, and the unit is meter/second;
when R belongs to [8000,9000), the sunshine semantic information is 'very low';
when R belongs to [9000,10000), the sunshine semantic information is 'low';
when R belongs to [10000,11000), the sunshine semantic information is 'general';
when R belongs to [11000,12000), the sunshine semantic information is 'high';
when R belongs to [12000,13000), the sunshine semantic information is 'very high';
wherein R is day sunshine and the unit is KJJ/m2；
When K belongs to [151,300], the environment semantic information is 'very low';
when K belongs to [131,150], the environment semantic information is 'low';
when K belongs to [101,130], the environment semantic information is 'general';
when K belongs to [51,100], the environment semantic information is 'high';
when K belongs to [0,50], the environment semantic information is 'very high';
wherein K is an air pollution index;
defining the power grid situation as no risk when the power grid is not in any fault;
when the power grid has a fault at the position 1 and power failure is not caused, defining the power grid situation as low risk;
when 1-3 faults occur in the power grid and power failure is not caused, defining the power grid situation as medium risk;
when the power grid has faults at or above 3 positions and does not cause power failure of the power grid, defining the power grid situation as high risk;
and when the power grid has faults at or above 3 positions and causes power failure of the power grid, defining the power grid situation as a fault state.
In the step 2, semantic information corresponding to weather, sunshine and environment historical data is converted into a one-dimensional array, namely fragile number, according to the following mode:
very low: (0, 0.1, 0.2), low: (0, 0.3, 0.5), typically (M): (0.5, 0.7, 0.9), high: (0.75, 0.85, 0.95), very high: (0.9, 0.95, 1.00);
converting the semantic information of the power grid situation into a one-dimensional array, namely a fragile number, according to the following mode:
the power grid situation is risk-free: (0, 0.1, 0.2), the grid situation is low risk: (0, 0.3, 0.5), the grid situation is at risk: (0.5, 0.7, 0.9), the grid situation is high risk: (0.75, 0.85, 0.95), the power grid situation is a fault state: (0.9,0.95,1.00).
In the step 2, the converted brittleness number is defuzzified, and the specific conversion mode is as follows:
Figure BDA0002835137740000041
wherein
Figure BDA0002835137740000042
Three values, f, corresponding to the triangular fuzzy numberi(t) is the value of the ith class data at time t.
In the step 3, the fuzzified brittleness number needs to be normalized, and the specific conversion mode is as follows:
Figure BDA0002835137740000043
wherein f isi(t) is the crispness number after defuzzification,
Figure BDA0002835137740000044
is a normalized value, fi maxIs the maximum value of the i-th class data.
In the step 4, the brittleness number obtained by converting the power grid operation data and the non-power grid operation data is used as two multidimensional arrays, and the normalized data is subjected to feature extraction through the following formulas respectively:
Figure BDA0002835137740000045
wherein, gcRepresenting the center point data of each layer in the multi-dimensional data, giRepresenting data around a center point of each layer of the multi-dimensional data, P representing a total amount of data around the center point of each layer of the multi-dimensional data,
Figure BDA0002835137740000046
for the finally obtained feature data, s is a piecewise function and x is (g)i-gc) The value of (c).
In the step 5, training the characteristic data through a BP neural network, wherein the input quantity of the input layer node is the characteristic data, and a power grid situation prediction model is obtained by training the characteristic data; the BP neural network is divided into an input layer, a hidden layer and an output layer; the data of the input layer is the characteristic data extracted in the step 4; values of hidden layer nodes j
Figure BDA0002835137740000047
Is an input value of the output layer, wherein wijFor weights of input layer to hidden layer, xiIs the input value of the input layer, and n is the number of nodes of the input layer; the node of the output layer has
Figure BDA0002835137740000051
Wherein i, j respectively represent a certain node of the input layer, the hidden layer and the output layer, vjThe weight from the node j in the hidden layer to the output layer, and the number of nodes in the hidden layer.
And 5, training the characteristic data by using the BP neural network, wherein the error between the known power grid situation and the predicted power grid situation of the BP neural network is ErAdjusting variable of connection weight
Figure BDA0002835137740000052
And wijIs adjusted variable of
Figure BDA0002835137740000053
Are respectively shown in the following formulas, wherein wijFor the weight of input layer node i to hidden layer node j:
Figure BDA0002835137740000054
Figure BDA0002835137740000055
Figure BDA0002835137740000056
wherein d isrIs the known grid situation, yrIs the power grid situation predicted by BP neural network, eta is gain factor, and [0,1 ] is taken]A positive number of intervals.
In the step 5, the number of iterations for training the feature data by using the BP neural network is 20.
The invention also discloses a power grid operation characteristic sensing system based on the LBP and the neural network, which comprises a data acquisition module, a data preprocessing module, a characteristic extraction module and a power grid situation prediction module;
the data acquisition module acquires power grid operation historical data and non-power grid operation historical data; the power grid operation historical data comprises average voltage, average transmission power and power grid repair rate; the non-power grid operation historical data comprises historical data of weather, sunshine and environment at corresponding acquisition moments;
the data preprocessing module is used for representing collected non-power grid operation data including weather, sunshine, environment historical data and power grid situations corresponding to historical days through semantic information respectively, converting the semantic information into crisp numbers through triangular fuzzy numbers, and performing normalization processing on the power grid operation data and the non-power grid operation data;
the feature extraction module adopts an LBP method to extract features of the normalized multidimensional data;
the power grid situation prediction module trains the characteristic data through a BP neural network, the input quantity of the input layer node is the characteristic data, a power grid situation prediction model is obtained through training the characteristic data, and then the power grid situation is predicted.
The LBP and neural network-based power grid operation characteristic sensing method and system can be used for accurately extracting characteristics of power grid data and non-power grid data, training the characteristic data by using the BP neural network, acquiring the power grid situation, providing a basis for a dispatcher to sense the power grid state, and guaranteeing safe and reliable operation of a power grid.
Drawings
FIG. 1 is a flow chart of a power grid operation characteristic sensing method and system based on LBP and neural network;
FIG. 2 is a process of feature extraction for normalized data using LBP method;
FIG. 3 is a concrete model construction of a BP neural network;
fig. 4 is a system diagram of a power grid operation data feature extraction method based on LBP and neural network.
Detailed Description
The present application is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present application is not limited thereby.
As shown in fig. 1, the invention discloses a power grid operation characteristic sensing method and system based on lbp (local Binary patterns) and a neural network, comprising the following steps:
step 1, collecting power grid operation historical data, non-power grid operation historical data and power grid situation historical information; the power grid operation historical data comprise average voltage, average transmission power and power grid repair rate; the non-power grid operation historical data comprises historical data of weather, sunshine and environment at corresponding acquisition moments.
The weather historical data refers to the evaluation result of the weather wind speed, and the weather historical data respectively represent different weather conditions by adopting 'very high', 'normal', 'low' and 'very low' in order to facilitate calculation. The weather historical data comprises weather wind speed, the sunshine historical data refers to the intensity of solar radiation, and the environment historical data refers to historical air pollution indexes. When V belongs to [0,3.4), the weather semantic information is 'very low'; when V belongs to [3.4,8.0), the weather semantic information is 'low'; when V belongs to [8.0,10.7), the weather semantic information is 'general'; when V belongs to [10.8,13.8), the weather semantic information is 'high'; when V is larger than or equal to 13.8, the weather semantic information is 'very high'; wherein V is the current day wind speed, and the unit is meter/second; when R belongs to [8000,9000), the sunshine semantic information is 'very low'; when R belongs to [9000,10000), the sunshine semantic information is 'low'; when R belongs to [10000,11000), the sunshine semantic information is 'general'; when R belongs to [11000,12000), the sunshine semantic information is 'high'; when R belongs to [12000,13000), the sunshine semantic information is 'very high'; wherein R is day sunshine and the unit is KJJ/m2(ii) a When K ∈ [151,300]]The environment semantic information is 'very low'; when K ∈ [131,150]]Environment semantic information is "low"; when K ∈ [101,130]]Environment semantic information is 'general'; when K ∈ [51,100]]Environment semantic information is "high"; when K is equal to 0,50]Environment semantic information is "very high"; wherein K is an air pollution index;
the historical information of the power grid situation comprises power grid fault situations which are divided into five situations of no risk, low risk, medium risk, high risk and fault state. Defining the power grid situation as no risk when the power grid is not in any fault; when the power grid has a fault at the position 1 and power failure is not caused, defining the power grid situation as low risk; when 1-3 faults occur in the power grid and power failure is not caused, defining the power grid situation as medium risk; when the power grid has faults at or above 3 positions and does not cause power failure of the power grid, defining the power grid situation as high risk; and when the power grid has faults at or above 3 positions and causes power failure of the power grid, defining the power grid situation as a fault state.
The invention represents the three types of non-grid operation historical data of weather, sunshine and environment through semantics, the semantic information can be divided into multiple stages, the more the grade division is, the higher the precision is, and the higher the cost to be trained by a neural network is, the invention preferably selects 5 standards of 'very high', 'general', 'low' and 'very low', and technicians in the field can divide the semantic information according to actual requirements.
The numerical range standard for dividing the non-grid operation historical data is only a preferred embodiment, and a person skilled in the art should determine a proper division range according to local climate conditions to perform reasonable division.
And 2, converting the semantic information into fragile numbers through the triangular fuzzy numbers according to the table 1, and defuzzifying the converted triangular fuzzy numbers by adopting the following formula.
TABLE 1 triangular fuzzy number conversion table
Figure BDA0002835137740000071
Figure BDA0002835137740000072
Wherein
Figure BDA0002835137740000073
Three values, f, corresponding to the triangular fuzzy numberi(t) is the value of the ith class data at time t.
Step 3, normalizing the power grid data and the converted power grid data by using a formula (2):
Figure BDA0002835137740000081
wherein
Figure BDA0002835137740000082
Is a normalized value, fi maxIs the maximum value of the i-th class data.
Step 4, by using an LBP method, respectively taking fragile numbers obtained by converting power grid operation data and non-power grid operation data as multidimensional arrays, and performing feature extraction on the normalized data through the following formula, wherein the specific method is as shown in FIG. 2:
Figure BDA0002835137740000083
wherein, gcRepresenting the center point data of each layer in the multi-dimensional data, giRepresenting data around a center point of each layer of the multi-dimensional data, P representing a total amount of data around the center point of each layer of the multi-dimensional data,
Figure BDA0002835137740000084
for the finally obtained feature data, s is a piecewise function and x is (g)i-gc) The value of (c).
And 5, taking the obtained characteristic data as input and the power grid situation data as output, and training a power grid situation prediction model by using a BP neural network. The specific model of the BP neural network is constructed as shown in FIG. 3, wherein i, j respectively represent a certain node of the input layer, the hidden layer and the output layer, and n, m respectively represent the number of nodes of the input layer and the hidden layer. And 4, inputting the input quantity of the input layer node into the power grid characteristic data and the non-power grid characteristic data extracted in the step 4. For the hidden layer node j, there are:
Figure BDA0002835137740000085
wherein h isjAs input values to the output layer, w, for specific values of the nodes of the hidden layerijFor weights of input layer to hidden layer, xiIs the input value of the input layer, and n is the number of nodes of the input layer;
for the output nodes there are:
Figure BDA0002835137740000086
i, j respectively represent a certain node of the input layer, the hidden layer and the output layer, vjThe weights from the node j in the hidden layer to the output layer are shown, and n and m respectively represent the number of nodes of the input layer and the hidden layer.
The error of the known power grid situation and the predicted power grid situation of the BP neural network in the training process is ErAdjusting variable of connection weight
Figure BDA0002835137740000091
And wijIs adjusted variable of
Figure BDA0002835137740000092
Are respectively shown in the following formulas, wherein wijFor the weight of input layer node i to hidden layer node j:
Figure BDA0002835137740000093
Figure BDA0002835137740000094
Figure BDA0002835137740000095
wherein d isrIs the known grid situation, yrIs the predicted grid situation by the BP neural network. Eta is a gain factor, typically taken as [0,1 ]]A positive number of intervals.
And finally obtaining a prediction model through 20 iterations.
And 6, collecting power grid operation data and weather, sunshine and environment data, inputting the data into the power grid situation prediction model trained in the step 5, and predicting the power grid situation.
The invention also discloses a power grid operation characteristic perception system based on LBP and neural network as shown in figure 4, which comprises a data acquisition module, a data preprocessing module, a characteristic extraction module and a power grid situation prediction module:
the data acquisition module acquires power grid operation historical data and non-power grid operation historical data; the power grid operation historical data comprise average voltage, average transmission power and power grid repair rate; the non-power grid operation historical data comprises historical data of weather, sunshine and environment at corresponding acquisition moments;
the data preprocessing module is used for representing collected non-power grid operation data including weather, sunshine, environment historical data and power grid situations corresponding to historical days through semantic information respectively, converting the semantic information into crisp numbers through triangular fuzzy numbers, and performing normalization processing on the power grid operation data and the non-power grid operation data;
the feature extraction module adopts an LBP method to extract features of the normalized multidimensional data;
the power grid situation prediction module trains the characteristic data through a BP neural network, the input quantity of the input layer node is the characteristic data, a power grid situation prediction model is obtained through training the characteristic data, and then the power grid situation is predicted.
And then, extracting the characteristics of the actual operation data and non-power grid operation data of the power grid all the year around in a certain area of Suzhou city in Jiangsu province, inputting the data into a BP neural network for training, and obtaining a power grid situation prediction model. The embodiments are described herein.
Table 2 partial raw data
Figure BDA0002835137740000101
According to the contents of table 1, the weather, sunshine and environment semantic information is converted into fragile number, and the conversion result of part of the original data is shown in table 3:
TABLE 3 partial original semantic information conversion results
Figure BDA0002835137740000102
The data were normalized and the results of partial data normalization are shown in table 4:
table 4 partial raw data normalization results
Figure BDA0002835137740000103
And (3) performing feature extraction on the normalized data by adopting an LBP (local binary pattern) method, and respectively performing feature extraction on the power grid operation data and the non-power grid operation data in order to guarantee the effectiveness of the feature extraction. The final extraction results are shown in table 5:
table 5 data feature extraction results
Figure BDA0002835137740000111
And then, the acquired characteristic data is input, the power grid situation information is used as output to train the BP neural network, and finally, a final result is obtained through 20 iterations. Through verification, the accuracy of the obtained model can reach over 90 percent, and support can be provided for power grid dispatching personnel.
The LBP and neural network-based power grid operation characteristic sensing method and system can be used for accurately extracting characteristics of power grid data and non-power grid data, training the characteristic data by using the BP neural network, acquiring the power grid situation, providing a basis for a dispatcher to sense the power grid state, and guaranteeing safe and reliable operation of a power grid.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention. The present applicant has described and illustrated embodiments of the present invention in detail with reference to the accompanying drawings, but it should be understood by those skilled in the art that the above embodiments are merely preferred embodiments of the present invention, and the detailed description is only for the purpose of helping the reader to better understand the spirit of the present invention, and not for limiting the scope of the present invention, and on the contrary, any improvement or modification made based on the spirit of the present invention should fall within the scope of the present invention.

Claims (11)

1. A power grid situation perception method based on LBP and a neural network is characterized by comprising the following steps:
step 1: collecting power grid operation historical data, non-power grid operation historical data and power grid situation data; the power grid operation historical data comprises average voltage, average transmission power and power grid repair rate; the non-power grid operation historical data comprises historical data of weather, sunshine and environment at corresponding acquisition moments; the power grid situation data refer to power grid fault conditions and risk levels;
step 2: representing the non-power grid operation historical data and the power grid situation of the corresponding historical day through semantic information respectively, and converting the semantic information representing different risk levels of the non-power grid operation historical data range and the power grid situation into corresponding one-dimensional arrays, namely fragile numbers respectively;
and step 3: respectively carrying out normalization processing on the collected power grid operation data and the brittleness number converted in the step 2;
and 4, step 4: performing feature extraction on the power grid operation data and the brittleness number normalized in the step 3 to obtain feature data of power grid operation and feature data of non-power grid operation, and expressing the feature data in a numerical value form;
and 5: inputting the characteristic data extracted in the step (4) into a BP neural network for training to obtain a power grid situation prediction model;
step 6: and (5) collecting power grid operation data and non-power grid operation historical data, inputting the power grid operation data and the non-power grid operation historical data into the power grid situation prediction model trained in the step 5, and predicting the power grid situation.
2. The power grid operation characteristic sensing method based on the LBP and the neural network as claimed in claim 1, wherein:
in the step 1, collected power grid operation historical data is calendar historical data, the repair rate is the repair rate of the current day of the power grid, and the power grid situation is determined based on the repair rate of the current day of the power grid and the fault degree of the power grid;
the weather historical data refers to weather wind speed;
the sunshine history data refers to the intensity of solar radiation;
the environmental historical data specifically refers to historical air pollution indexes;
representing the power grid situation into five conditions of no risk, low risk, medium risk, high risk and fault state through semantic information;
and respectively representing the weather historical data, the sunshine historical data and the environment historical data as five conditions of high, common, low and low through semantic information.
3. The power grid operation characteristic sensing method based on the LBP and the neural network as claimed in claim 2, wherein:
the semantic information of the weather historical data is characterized in the following way:
when V belongs to [0,3.4), the weather semantic information is 'very low';
when V belongs to [3.4,8.0), the weather semantic information is 'low';
when V belongs to [8.0,10.7), the weather semantic information is 'general';
when V belongs to [10.8,13.8), the weather semantic information is 'high';
when V is larger than or equal to 13.8, the weather semantic information is 'very high';
wherein V is the current day wind speed, and the unit is meter/second;
when R belongs to [8000,9000), the sunshine semantic information is 'very low';
when R belongs to [9000,10000), the sunshine semantic information is 'low';
when R belongs to [10000,11000), the sunshine semantic information is 'general';
when R belongs to [11000,12000), the sunshine semantic information is 'high';
when R belongs to [12000,13000), the sunshine semantic information is 'very high';
wherein R is day sunshine and the unit is KJJ/m2；
When K belongs to [151,300], the environment semantic information is 'very low';
when K belongs to [131,150], the environment semantic information is 'low';
when K belongs to [101,130], the environment semantic information is 'general';
when K belongs to [51,100], the environment semantic information is 'high';
when K belongs to [0,50], the environment semantic information is 'very high';
wherein K is an air pollution index;
defining the power grid situation as no risk when the power grid is not in any fault;
when the power grid has a fault at the position 1 and power failure is not caused, defining the power grid situation as low risk;
when 1-3 faults occur in the power grid and power failure is not caused, defining the power grid situation as medium risk;
when the power grid has faults at or above 3 positions and does not cause power failure of the power grid, defining the power grid situation as high risk;
and when the power grid has faults at or above 3 positions and causes power failure of the power grid, defining the power grid situation as a fault state.
4. The power grid operation characteristic sensing method based on the LBP and the neural network as claimed in claim 3, wherein:
in the step 2, semantic information corresponding to weather, sunshine and environment historical data is converted into a one-dimensional array, namely fragile number, according to the following mode:
very low: (0, 0.1, 0.2), low: (0, 0.3, 0.5), typically (M): (0.5, 0.7, 0.9), high: (0.75, 0.85, 0.95), very high: (0.9, 0.95, 1.00);
converting the semantic information of the power grid situation into a one-dimensional array, namely a fragile number, according to the following mode:
the power grid situation is risk-free: (0, 0.1, 0.2), the grid situation is low risk: (0, 0.3, 0.5), the grid situation is at risk: (0.5, 0.7, 0.9), the grid situation is high risk: (0.75, 0.85, 0.95), the power grid situation is a fault state: (0.9,0.95,1.00).
5. The power grid operation characteristic sensing method based on the LBP and the neural network as claimed in claim 2 or 4, wherein:
in the step 2, the converted brittleness number is defuzzified, and the specific conversion mode is as follows:
Figure FDA0002835137730000031
wherein
Figure FDA0002835137730000032
Three values, f, corresponding to the triangular fuzzy numberi(t) is the value of the ith class data at time t.
6. The power grid operation characteristic sensing method based on the LBP and the neural network as claimed in claim 5, wherein:
in the step 3, the fuzzified brittleness number needs to be normalized, and the specific conversion mode is as follows:
Figure FDA0002835137730000033
wherein f isi(t) is the crispness number after defuzzification,
Figure FDA0002835137730000034
is a normalized value, fi maxIs the maximum value of the i-th class data.
7. The power grid operation feature perception method based on the LBP and the neural network as claimed in claim 6, wherein:
in the step 4, the brittleness number obtained by converting the power grid operation data and the non-power grid operation data is used as two multidimensional arrays, and the normalized data is subjected to feature extraction through the following formulas respectively:
Figure FDA0002835137730000041
wherein, gcRepresenting the center point data of each layer in the multi-dimensional data, giRepresenting data around a center point of each layer of the multi-dimensional data, P representing a total amount of data around the center point of each layer of the multi-dimensional data,
Figure FDA0002835137730000042
for the finally obtained feature data, s is a piecewise function and x is (g)i-gc) The value of (c).
8. The LBP and neural network based power grid operation feature awareness method according to claim 1 or 7, wherein:
in the step 5, training the characteristic data through a BP neural network, wherein the input quantity of the input layer node is the characteristic data, and a power grid situation prediction model is obtained by training the characteristic data; the BP neural network is divided into an input layer, a hidden layer and an output layer; the data of the input layer is the characteristic data extracted in the step 4; values of hidden layer nodes j
Figure FDA0002835137730000043
j is 0,1,2,3 is the input value of the output layer, wijFor weights of input layer to hidden layer, xiIs the input value of the input layer, and n is the number of nodes of the input layer; the node of the output layer has
Figure FDA0002835137730000044
Wherein i, j represent the input layer,A certain node of the hidden layer or the output layer, vjThe weight from the node j in the hidden layer to the output layer, and the number of nodes in the hidden layer.
9. The LBP and neural network-based grid operation feature awareness method according to claim 8, wherein:
and 5, training the characteristic data by using the BP neural network, wherein the error between the known power grid situation and the predicted power grid situation of the BP neural network is ErAdjusting variable of connection weight
Figure FDA0002835137730000048
And wijIs adjusted variable of
Figure FDA0002835137730000045
Are respectively shown in the following formulas, wherein wijFor the weight of input layer node i to hidden layer node j:
Figure FDA0002835137730000046
Figure FDA0002835137730000047
Figure FDA0002835137730000051
wherein d isrIs the known grid situation, yrIs the power grid situation predicted by BP neural network, eta is gain factor, and [0,1 ] is taken]A positive number of intervals.
10. The LBP and neural network based grid operation feature awareness method according to claim 1 or 9, wherein: in the step 5, the number of iterations for training the feature data by using the BP neural network is 20.
11. The utility model provides a power grid operation characteristic perception system based on LBP and neural network, includes data acquisition module, data preprocessing module, feature extraction module and power grid situation prediction module, its characterized in that:
the data acquisition module acquires power grid operation historical data and non-power grid operation historical data; the power grid operation historical data comprises average voltage, average transmission power and power grid repair rate; the non-power grid operation historical data comprises historical data of weather, sunshine and environment at corresponding acquisition moments;
the data preprocessing module is used for representing collected non-power grid operation data including weather, sunshine, environment historical data and power grid situations corresponding to historical days through semantic information respectively, converting the semantic information into crisp numbers through triangular fuzzy numbers, and performing normalization processing on the power grid operation data and the non-power grid operation data;
the feature extraction module adopts an LBP method to extract features of the normalized multidimensional data;
the power grid situation prediction module trains the characteristic data through a BP neural network, the input quantity of the input layer node is the characteristic data, a power grid situation prediction model is obtained through training the characteristic data, and then the power grid situation is predicted.
CN202011467995.4A 2020-12-14 2020-12-14 Power grid operation characteristic sensing method and system based on LBP and neural network Active CN112488417B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011467995.4A CN112488417B (en) 2020-12-14 2020-12-14 Power grid operation characteristic sensing method and system based on LBP and neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011467995.4A CN112488417B (en) 2020-12-14 2020-12-14 Power grid operation characteristic sensing method and system based on LBP and neural network

Publications (2)

Publication Number Publication Date
CN112488417A true CN112488417A (en) 2021-03-12
CN112488417B CN112488417B (en) 2022-06-21

Family

ID=74917600

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011467995.4A Active CN112488417B (en) 2020-12-14 2020-12-14 Power grid operation characteristic sensing method and system based on LBP and neural network

Country Status (1)

Country Link
CN (1) CN112488417B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105631558A (en) * 2016-03-22 2016-06-01 国家电网公司 BP neural network photovoltaic power generation system power prediction method based on similar day
CN109193652A (en) * 2018-10-30 2019-01-11 贵州电网有限责任公司 A kind of distribution network failure self-healing system containing distributed generation resource based on Situation Awareness
CN110380897A (en) * 2019-07-04 2019-10-25 湖北央中巨石信息技术有限公司 Network security situation awareness model and method based on improved BP
CN111144638A (en) * 2019-12-24 2020-05-12 东南大学 Power distribution network operation situation prediction method based on big data
CN111582571A (en) * 2020-04-30 2020-08-25 中国电力科学研究院有限公司 Power grid operation situation sensing method and system with model driving and data driving integrated
CN112070215A (en) * 2020-09-10 2020-12-11 北京理工大学 BP neural network-based dangerous situation analysis processing method and processing device

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105631558A (en) * 2016-03-22 2016-06-01 国家电网公司 BP neural network photovoltaic power generation system power prediction method based on similar day
CN109193652A (en) * 2018-10-30 2019-01-11 贵州电网有限责任公司 A kind of distribution network failure self-healing system containing distributed generation resource based on Situation Awareness
CN110380897A (en) * 2019-07-04 2019-10-25 湖北央中巨石信息技术有限公司 Network security situation awareness model and method based on improved BP
CN111144638A (en) * 2019-12-24 2020-05-12 东南大学 Power distribution network operation situation prediction method based on big data
CN111582571A (en) * 2020-04-30 2020-08-25 中国电力科学研究院有限公司 Power grid operation situation sensing method and system with model driving and data driving integrated
CN112070215A (en) * 2020-09-10 2020-12-11 北京理工大学 BP neural network-based dangerous situation analysis processing method and processing device

Also Published As

Publication number Publication date
CN112488417B (en) 2022-06-21

Similar Documents

Publication Publication Date Title
CN109546659B (en) Power distribution network reactive power optimization method based on random matrix and intelligent scene matching
CN111753893A (en) Wind turbine generator power cluster prediction method based on clustering and deep learning
CN110490385A (en) The unified prediction of electric load and thermic load in a kind of integrated energy system
CN104573879A (en) Photovoltaic power station output predicting method based on optimal similar day set
CN102102626A (en) Method for forecasting short-term power in wind power station
Zhu et al. A prediction model for wind farm power generation based on fuzzy modeling
CN110516840A (en) Short term prediction method based on the wind light generation power output for improving random forest method
CN111488896B (en) Distribution line time-varying fault probability calculation method based on multi-source data mining
CN111832812A (en) Wind power short-term prediction method based on deep learning
WO2023201552A1 (en) County-wide photovoltaic prediction method based on cluster division and data enhancement
CN113657662B (en) Downscaling wind power prediction method based on data fusion
CN103049609B (en) Wind power multi-stage scene simulation method
CN115829105A (en) Photovoltaic power prediction method based on historical data feature search
CN109272258B (en) Regional wind and solar power generation resource evaluation method based on K-means clustering
CN111209934A (en) Fan fault prediction and alarm method and system
CN105825002A (en) Method for modeling dynamic equivalence of wind power farm based on dynamic grey-relevancy analysis method
CN114399081A (en) Photovoltaic power generation power prediction method based on weather classification
CN113516271A (en) Wind power cluster power day-ahead prediction method based on space-time neural network
CN112288157A (en) Wind power plant power prediction method based on fuzzy clustering and deep reinforcement learning
CN113449847B (en) Offshore wind power rolling prediction method considering second-level time series wind speed change
CN113610285A (en) Power prediction method for distributed wind power
CN112488417B (en) Power grid operation characteristic sensing method and system based on LBP and neural network
CN115660032B (en) BI-LSTM neural network integration attention mechanism-based building roof photovoltaic power prediction method
Guo et al. Wind speed forecasting of genetic neural model based on rough set theory
CN112039051A (en) Real-time modeling method for accessing double-fed wind driven generator into substation bus load

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant